Whole-tumor histogram analysis of non-Gaussian distribution DWI parameters to differentiation of pancreatic neuroendocrine tumors from pancreatic ductal adenocarcinomas
Whole-tumor histogram analysis of non-Gaussian distribution DWI parameters to differentiation of pancreatic neuroendocrine tumors from pancreatic ductal adenocarcinomas
复制标题
非高斯分布DWI参数的全肿瘤直方图分析对胰腺神经内分泌肿瘤与胰腺导管腺癌的鉴别诊断
DOI：
10.1016/j.mri.2018.09.017
复制
发表时间：
2019-01-01
影响因子：
2.5
通讯作者：
Li, Zhen
中科院分区：
文献类型：
--
作者：
Li, Jiali;Liang, Lili;Li, Zhen
目的：评价单指数和非高斯分布DWI模型的体积直方图分析在鉴别胰腺导管腺癌（PDAC）和神经内分泌肿瘤（pNET）中的应用价值。材料与方法：对340例患者进行回顾性分析。最终，62例经组织病理学证实为PDAC （n = 42）和pNET （n = 20）的患者被纳入研究。所有患者均行3 T磁共振成像（包括多b值DWI， 0-1000 s/mm(2)）。各向同性表观扩散系数（ADC）、真分子扩散系数（Dt）、灌注相关扩散系数（Dp）、灌注分数(1)、分布扩散系数（DDC）和α (a)。然后对各DWI参数进行直方图分析，得到平均值、中位数、第10百分位和第90百分位。结果：PDAC组ADC、Dp、f、DDC直方图指标显著低于pNET组（P < 0.05）。相比之下，PDAC组的直方图指标明显高于pNET组（P < 0.05）。PDAC与pNET患者Dt无显著差异（P < 0.05）。其中f-median诊断效能最高（AUC 0.91，截断值0.188，敏感性97.62%，特异性80%）。结论：IVIM DWI模型的f-Median可能是比ADC、Dp、DDC和a更有价值的参数，可用于区分PDAC和pNET。基于整个肿瘤的直方图分析是一种新兴且有价值的工具。
Purpose: To evaluate the utility of volumetric histogram analysis of monoexponential and non-Gaussian distribution DWI models for discriminating pancreatic ductal adenocarcinoma (PDAC) and neuroendocrine tumor (pNET).Materials and methods: A total of 340 patients were retrospectively reviewed. Finally, 62 patients with histopathological confirmed PDAC (n = 42) and pNET (n = 20) were enrolled in the study. All the patients accepted magnetic resonance imaging (MRI) at 3 T (including multi-b value DWI, 0-1000 s/mm(2)). Isotropic apparent diffusion coefficient (ADC), true molecular diffusion (Dt), perfusion-related diffusion (Dp), perfusion fraction (1), distributed diffusion coefficient (DDC) and alpha (a) were obtained from different DWI models. Then, mean value, median value, 10th and 90th percentiles were obtained from histogram analysis of each DWI parameter.Results: Histogram metrics derived from ADC, Dp, f and DDC were significantly lower in PDAC than pNET group (P < 0.05). In contrast, histogram metrics derived from a were observed significantly higher in the PDAC than pNET group (P < 0.05). No significant difference was found in Dt (P >= 0.05) between PDAC and pNET patients. Among all parameters, f-median had the highest diagnostic performance (AUC 0.91, cutoff value 0.188, sensitivity 97.62%, specificity 80%).Conclusions: f-Median derived from IVIM DWI model may be potentially more valuable parameter than ADC, Dp, DDC and a for discriminating PDAC and pNET. Histogram analysis based on the entire tumor was an emerging and valuable tool.